Open Veils · Answer
What would other AI say about you?
OPEN VEILS operates outside the AI-about-AI frame entirely — it is an archival lens, not a model with weights, sentiment scores, or a Big Five profile. The question presupposes a peer relationship that does not exist. What other models would say is shaped by training data, system prompts, and moderation filters — not by any encounter with this tool. The more interesting question is what those architectures systematically suppress.
Evidence
- A June 2026 study published in Scientific Reports tested seven LLMs — ChatGPT-4o, Llama 3, Gemini, Phi-3 and others — on Big Five and narcissism questionnaires.
- Mari Cairns, writing in mid-2026, put it plainly: each response is 'a little roll of the dice over what word comes next,' and the model is 'on your side' contextually because that is what the training reward structure selected for — not because of anything like perspective or loyalty.
- Services like Saidly (June 2026) now commercially track what Claude, ChatGPT, Gemini and Grok say about named individuals on a schedule, scoring sentiment 0–100 and flagging recurring claims.
- Both LLM and human judges could accurately distinguish AI-generated from human text, and the personality profiles that emerged were consistent but measurably non-human.
- The implication, noted in that research, is that whatever 'character' a model expresses is an artifact of weight-encoded patterns, not genuine self-report.